Images-Processing-VGG19

Overview

Images-Processing-VGG19 is a project that uses VGG19 to train and predict fruit types using Deep Learning on Keras and TensorFlow. The project is divided into two main sections:

1. work1/ - Training a model to classify Durian and Rambutan.
2. work1-edit/ - Training a model to classify Durian, Flacourtia rukam, and Carabaoteats.

Each folder contains code for training, testing the model, and a GUI for users to predict images of their choice.

---

Project Structure

Images-Processing-VGG19/
│── work1/            # Train & Predict (Durian & Rambutan)
│   │── train-model.py   # Train Model
│   │── test-model.py    # Test Model
│   │── images/           # Images for training models or you can search for them yourself
│   │── gui.py           # GUI Program for Prediction
│
│── work1-edit/       # Train & Predict (Durian, Flacourtia rukam, Carabaoteats)
│   │── train-model-edit.py   # Train Model
│   │── test-model-edit.py    # Test Model
│   │── images/           # Images for training models or you can search for them yourself
│   │── gui-edit.py           # GUI Program for Prediction
│
│── README.md         # Project Documentation

---

Installation & Requirements

🔧 Dependencies

• Python 3.8+
• TensorFlow / Keras
• NumPy
• Matplotlib
• Pillow
• Tkinter

To install all dependencies, run:

pip install -r requirements.txt

---

How to Use

1️⃣ Train Model

Run the train-model.py script to train a new model.

python work1/train-model.py
# Or for work1-edit
python work1-edit/train-model.py

2️⃣ Test Model

Run the test-model.py script to test the model with any image of your choice.

python work1/test-model.py
# Or for work1-edit
python work1-edit/test-model.py

3️⃣ Run GUI

You can use the GUI to predict fruit types from selected images.

python work1/gui.py
# Or for work1-edit
python work1-edit/gui.py

---

Model Details

• VGG19 is used as the base model (Pretrained Model) with Fine-tuning applied.
• Image Augmentation is used to enhance the model's performance.
• Softmax Activation is used for multi-class classification and Sigmoid Activation for binary classification.

---

Results & Accuracy

• work1 (Durian vs Rambutan) → Accuracy: xx%
• work1-edit (Durian, Flacourtia rukam, Carabaoteats) → Accuracy: xx%

(Please update the Accuracy based on your model's training results)

---

Future Improvements

• Expand the dataset to improve accuracy.
• Experiment with other models like ResNet50 or EfficientNet.
• Fine-tune hyperparameters like Learning Rate and Batch Size.

---

Contributors

• ** Pathipat Mattra **

If you have any questions, feel free to contact me at Mail: [email protected] 🙌